Can Sand Dunes Be Used to Study Historic Storm Events?

This is a repository copy of Can sand dunes be used to study historic storm events?.

White Rose Research Online URL for this paper: https://eprints.whiterose.ac.uk/124685/

Version: Published Version

Article: Bateman, Mark D., Rushby, Greg, Stein, Sam et al. (4 more authors) (2017) Can sand dunes be used to study historic storm events? EARTH SURFACE PROCESSES AND LANDFORMS. ISSN 0197-9337 https://doi.org/10.1002/esp.4255

Reuse This article is distributed under the terms of the Creative Commons Attribution (CC BY) licence. This licence allows you to distribute, remix, tweak, and build upon the work, even commercially, as long as you credit the authors for the original work. More information and the full terms of the licence here: https://creativecommons.org/licenses/

Takedown If you consider content in White Rose Research Online to be in breach of UK law, please notify us by emailing [email protected] including the URL of the record and the reason for the withdrawal request.

[email protected] https://eprints.whiterose.ac.uk/ EARTH SURFACE PROCESSES AND LANDFORMS Earth Surf. Process. Landforms (2017) © 2017 The Authors. Earth Surface Processes and Landforms published by John Wiley & Sons Ltd. Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/esp.4255

Can sand dunes be used to study historic storm events?

Mark D. Bateman,1* Greg Rushby,1 Sam Stein,1 Robert A. Ashurst,1 David Stevenson,1 Julie M. Jones1 and W. Roland Gehrels2 1 Geography Department, University of Sheffield, Winter St., Sheffield S10 2TN, UK 2 Environment Department, University of York, Heslington, York YO10 5NG, UK

Received 19 May 2017; Revised 5 September 2017; Accepted 5 September 2017

*Correspondence to: Mark D. Bateman, Geography Department, University of Sheffield, Winter St., Sheffield S10 2TN, UK. E-mail: [email protected] This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

ABSTRACT: Knowing the long-term frequency of high magnitude storm events that cause coastal inundation is critical for present coastal management, especially in the context of rising sea levels and potentially increasing frequency and severity of storm events. Coastal sand dunes may provide a sedimentary archive of past storm events from which long-term frequencies of large storms can be reconstructed. This study uses novel portable optically stimulated luminescence (POSL) profiles from coastal dunes to reconstruct the sedimentary archive of storm and surge activity for Norfolk, UK. Application of POSL profiling with supporting luminescence ages and particle size analysis to coastal dunes provides not only information of dunefield evolution but also on past coastal storms. In this study, seven storm events, two major, were identified from the dune archive spanning the last 140 years. These appear to correspond to historical reports of major storm surges. Dunes appear to be only recording (at least at the sampling resolution used here) the highest storm levels that were associated with significant flooding. As such the approach seems to hold promise to obtain a better understanding of the frequency of large storms by extending the dune archive records further back to times when documentation of storm surges was sparse. © 2017 The Authors. Earth Surface Processes and Landforms published by John Wiley & Sons Ltd.

KEYWORDS: dunes; particle size; portable OSL; Norfolk; storms

Introduction given rising mean sea levels, will be less for an equivalent surge in the near future (Wolf and Flather, 2005). While hydrological On the 5 December 2013, a deep depression north of Scotland, modelling allows better prediction of peak storm-tide height UK, in conjunction with spring tides, produced an extreme along coastlines, further refinement requires better understand- storm surge down the east coast of Britain (Sibley et al., ing of beach parameters and other coastal zone factors (Lewis 2015). During this water reached 6.05 m OD (Ordnance et al., 2013). To achieve this also requires an understanding Datum, which approximates mean sea level) at Kings Lynn of the response of specific segments of the coastline over longer (Spencer et al., 2015) and 6.31 m OD at Wells-Next-The-Sea time-scales (Andrews et al., 2000). Coastal sand dunes may in Norfolk causing extensive damage to coastal infrastructure, provide a sedimentary archive of past storm events. housing and farmland (Figure 1). This was the worst since Coastal dunes serve as an important buffer zone protecting the disastrous flooding of southern North Sea coasts in 1953 low-lying coastal areas, and their growth is a function of ac- (Baxter, 2005; Gerritsen, 2005) when several thousand deaths commodation space, sediment supply, wind and preservation. occurred across NW Europe (>300 on the east coast of the In particular wide beaches, providing a plentiful sediment UK), as well as extensive breaching of coastal flood defences supply, and a sequence of years without high-intensity storms (Spencer et al., 2015). Knowing the long-term frequency of promote dune development (Puijenbroek et al., 2017). During such high magnitude events as well as more frequent severe storm events they can be modified (Spencer et al., 2015). The storms is critical for present coastal management (Lewis et al., nature of this modification is partly a function of how well 2013) particularly for low-lying coastlines where small changes dunes have coalesced to form a perpendicular barrier to the in sea levels can translate into flooding a long way inland. It is sea, the height above OD of the dune crest and any low points also of increasing importance given rising sea levels (Church et within the dunes. If waves come into direct contact with dunes al., 2013) and the potential for future increased storminess over then erosion may take the form of breaching and overwash central Europe (Zappa et al., 2013). North Sea storm surge aprons (Figure 2(a); Steers et al., 1979). This is more likely to levels are thought to be going to be substantially higher and occur where dunes are poorly developed (i.e. non continuous more frequent by the end of this century (Vousdoukas et al., along the back beach or low in elevation) and/or the magnitude 2016). The return period for extreme water levels to that of of the storm exceeds the height of the dunes. Alternatively, the 1953 storm surge was forecast to be around 50 years but, scarping may occur in which the leading slope of a dune is M. D. BATEMAN ET AL.

Figure 1. North Norfolk coastal damage resulting from the December 2013 storm surge in the North Sea. (a) Dune scarping at Brancaster, note metal pilings used to limit erosion on beach. (b) Flotsam left on an artificial sea-wall. The tidal surge reached the top but did not breach the wall on this occasion. (c) Erosion behind gabions placed to protect dunes at Brancaster. (d) Piles of former beach huts on Wells-Next-The-Sea destroyed in the storm surge. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 2. Impacts on coastal dunes of Storm surges. (a) Breaching of dune by waves forming sediment fan in newly formed interdune low. (b) Dune erosion forming ‘scarped’ vertical sand cliffs. (c) The same dunes as shown in (b) but 12 months on with collapsed and newly deflated sediment piled up against dunes. (d) Thin sheet of coarse grained sand deposited by 2013 storm event on dune crest. [Colour figure can be viewed at wileyonlinelibrary.com] eroded away leaving a vertical face in the dune on the seaward Given the above, for dunes to record a record of previous side (Figure 2(b)). This vertical face in sand is unstable so after a low frequency high magnitude storms within their stratigraphy storm event it collapses until a slope at the angle of repose is they need to meet the folowing requirements: re-established (Figure 2(c)). Recovery can be quick (<5 years, Brooks et al., 2016). This is more likely to occur for well • form a continuous back beach barrier (or inland barrier); established dunes with high crestlines and/or during lower • have crest lines consistently higher than all but the most magnitude storm events in which waves are sufficiently high extreme storm events preventing overtopping and erosion to reach dune fronts but not to over top them. Storms can also away of the sediment archive; modify dunes through deflation and/or the deposition of • be in an area in which sediment supply is allowing active coarser aeolian sediments moved on high winds during the dune accretion thereby allowing dunes affected by storm storm (Figure 2(d)). It is these erosion layers and storm scarping to be restored and dune crests to build up. deposited sediments which raises the potential for actively accreting dunes to have recorded longer-term coastal storm Proof of concept work has been carried out using innovative records. new portable optically stimulated luminescence (POSL) to

© 2017 The Authors. Earth Surface Processes and Landforms published by John Wiley & Sons Ltd. Earth Surf. Process. Landforms, (2017) COASTAL SAND DUNES AND STORMS establish high resolution relative coastal sand dune chronolo- providing sediment to the ~6.5 km of well developed coastal gies (Bateman et al., 2015). Actual age determinations are not dunes (Figure 3). Oronsaye (1990) showed dune accumulation possible with this method but total POSL counts can be viewed was dominantly by wind with a northerly component. This as a proxy for age, with older samples having a higher lumines- makes the site idea for capturing the results of low frequency cence signal than younger samples. Unless the background high magnitude storms from the north such as the 1953 storm dose creating the luminescence signal changes through time surge rather than prevailing UK storms which are from the the POSL response with age should be linear. Down core dune south-west. Despite a backdrop of sea levels rising over the past À profiles showed a pattern of very similar POSL signals suddenly 7500 years on average by 0.6 ± 0.2 mm/a 1 (corrected for stepping up in signal (Bateman et al., 2015). These were changes in tidal range; Shennan and Horton, 2002), Brooks interpreted as similar aged sediment deposition through dune et al. (2016) showed that the coastline around Holkham Bay accretion with the steps indicating hiatuses in dune construction is prograding, increasing the likelihood of dune sediment caused by storms or tidal surges (ibid). This study seeks to preservation. A map from 1781 by Biedermann shows dunes examine coastal dune stratigraphy using POSL profiles from in existence at Holkham at this time (Hiskey, 2003) and based coastal dunes to determine if dunes act as sedimentary archives on historic maps and aerial imagery, dune formation has of storm and surge activity. continued to the present. The work of Bristow et al. (2000) at Holkham, using GPR, reported two items critical to the current study. First, it showed Region of study that the dunes have never been free dunes moving inland with time but always, once established, have been pinned by This study focuses on the Norfolk coastline near Holkham, vegetation responding only to erosion and accretion without ~20 km west of Hunstanton and 3 km east of Wells- much lateral movement. The stacking of dune sediment, rather Next-The-Sea. This is within the UK Environment Agency’s than overturning during movement, greatly enhances the ‘Superfrontage 2’ stretch of coastline for shoreline management potential for dunes to retain records of storm events and for purposes (East Anglian Coastal Group, 2010). This study area successful application of luminescence dating. Second, Bristow was chosen as the north-facing coastline has a long fetch and et al. (2000) were able to identify erosion surfaces extending a long history of being impacted by storms and storm surges. from the beach into the dunes which they attributed to creation It is fronted by broad sandy beaches and/or subtidal sand flats of dune cliffs subsequently covered by latter dune foreslope

Figure 3. The coastal area between Burnham Overy and Wells-Next-The-Sea including the sampled dunes sites of this study. Sites 1, 3 and 4 were on the foredunes found at the back of the present day beach. Sites 2 and 5 in inland dunes. Lower image modified from 2007 Digital Globe/Get Mapping image on Google Earth. [Colour figure can be viewed at wileyonlinelibrary.com]

© 2017 The Authors. Earth Surface Processes and Landforms published by John Wiley & Sons Ltd. Earth Surf. Process. Landforms, (2017) M. D. BATEMAN ET AL. activity. These erosion events were linked to documented storms, although their age was unknown. Three sets of dunes were selected for this study (Figure 3).

The first set of sites (Sites 1, 3 and 4) were located at three 1758 ± 20 < positions along the dunes found at the back of the present beach. The second set of dunes (Site 2) was from a line of dunes parallel to the coast, approximately 300 m inland, and interpreted as an older formation than the back-beach dunes (Figure 3). The final set (Site 5) was on a less distinct set of dunes between sets 1 and 2 (Figure 3). (Gy) n Age (years AD) e Methods

At all sample sites dune crest locations were selected and a )D

Dormer engineering sand drill used to core down to dune 1 base. The latter was established based on abney level surveys À of the dune height above beach and from a change in Gy ka μ sediment size from well-sorted medium sand to much coarser ( less sorted beach sand. At approximately 25 cm intervals, Total dose rate samples for POSL measurements and particle size analysis were collected in light-tight small plastic containers. Full ) optically stimulated luminescence (OSL) dating samples were 1 À also collected from near the dune surface, mid-core and base

of core at most sites. This resulted in the collection of a total Gy ka μ of 75 POSL and sediment samples and 12 samples for (

OSL dating. Cosmic dose rate For each POSL sample, any potentially light contaminated was removed and the sample material gently dried. All POSL measurements were undertaken with a SUERC portable OSL reader (Sanderson and Murphy, 2010) with measurements as per Bateman et al. (2015). Samples for full OSL dating were pre- pared as per Bateman and Catt, 1996), measured at the single aliquot level using a Risø luminescence reader and the single aliquot regeneration protocol (Murray and Wintle, 2003). Dose rates were determined from in situ field gamma- spectrometer measurements or elemental analysis. Results are shown in Table I (see Supplementary information for full details on the PSOL and OSL methods employed). All samples also underwent particle size analysis using a Horiba LA-950 laser diffraction particle size distribution analyser to detect changes in depositional environment. For this, sub-samples were riffled down and treated with 0.1% so- dium hexametaphosphate, before dispersal in de-ionised water within the instrument using ultrasound and pumping. Resultant data were used to calculate the mean and median grain size of each sample (Φ), as well as sorting, skewness, and probability distribution (Table II). Particle size showed nearly all samples to be well to moderately well sorted, unimodal with mean values in line with those found for dunes elsewhere (Lancaster, 2013). Within this were thin layers of slightly coarser and less well sorted sand which are interpreted as being aeolian storm deposits. As shown in Figure 4 particle size distributions from what are thought to be a storm deposit are slightly coarser, with a coarser tail and the 5th and 95th percentile further apart indicating poorer sorting. To provide a long-term context for the Holkham dunes two approaches were used. First, changes in coastline were reconstructed from historical maps (1st series Ordnance Survey), aerial imagery (Crown copyright Royal Air Force aerial photo- graphs 1946, Norfolk County Council aerial survey 1988) and Shfd13046Shfd14098Shfd14100 6.22Shfd14102 6.36 6.05 2.90 5.1 2.2 4.3 0.5 3.5 0.5 0.43 0.5 0.40 0.4 1.2 0.40 1.1 0.39 0.9 95 ± 5 1.0 94 ± 5 743 98 ± ± 34 5 144 ± 7 747 ± 46 734 ± 45 0.103 ± 0.014 687 ± 38 0.054 ± 24 0.013 0.107 ± 25 0.018 1874 ± 20 0.046 ± 0.014 17 1943 ± 18 22 1868 ± 26 1945 ± 20 Get Mapping imagery 2007 (via Google Earth). This resulted Shfd13042Shfd13043Shfd13045 3.68 5.93 3.40 3.6 4.3 2.1 0.5 0.5 0.5 0.34 0.46 0.43 1.0 1.1 1.1 129 ± 6 99 ± 5 134 ± 7 755 ± 35 753 ± 35 798 ± 36 0.034 ± 0.013 0.035 ± 0.013 24 0.203 ± 0.013 24 24 1968 ± 17 1966 ± 17 in 11 time slices spanning 175 years back to 1840.

The second approach was to reconstruct a coastal storm OSL related data for sampled sites within the Holkham dunes flooding record from reported observations and secondary documentary sources (Table III). Brooks et al. (2016) reported Site 3Site 4Site Shfd14097 5 Shfd14099 Shfd14101 1.00 1.00 0.84 3.2 2.9 3.9 0.5 0.4 0.38 0.5 0.37 1.0 0.31 1.0 0.9 185 ± 9 185 ± 9 818 188 ± ± 46 9 725 ± 38 796 ± 0.038 46 ± 0.014 0.013 ± 24 0.019 0.036 ± 13 0.015 1968 ± 17 21 1995 ± 30 1965 ± 20 Site 1 Shfd13041 0.87 3.3 0.5 0.4 1.1 187 ± 9 835 ± 34 0.027 ± 0.014 24 1981 ± 16 Site 2 Shfd13044 0.80 3.4 0.5 0.4 1.1 189 ± 9 836 36± 0.071 ± 0.013 24 1938 ± 16 Sample site Sample code Depth from surface (m) Water content (%) K (%) U (ppm) Th (ppm) 21 observed north Norfolk coastal storms from 1883 to 2013. Table I.

Table II. Particle size measurement results. Entries shaded yellow indicate samples >5% coarser and/or where sorting is >5% worse than the average for that dune. Lines indicate where OSL shows a temporal hiatus in the dune record. Entries in red are those identified as potential storm deposits

Depth Mean Sorting Skew Kurtosis Depth Mean Sorting Skew Kurtosis Dune Phase (m) (Mz) (σI) (SKI) (KG) Dune Phase (m) (Mz) (σI) (SKI) (KG) Site 1 (Back-beach dune) Site 3 (Back-beach dune) 1 0.64 1.70 -0.45 0.03 1.06 0.78 1.80 -0.47 0.04 1.08 0.87 1.82 -0.44 0.04 1.07 1.15 1.81 -0.46 0.05 1.08 0.96 1.79 -0.46 0.05 1.04 1.50 1.99 -0.43 0.07 1.10 2 1.35 1.94 -0.45 0.10 1.08 1.82 1.77 -0.45 0.05 1.05 1.70 1.93 -0.44 0.07 1.06 2.16 1.86 -0.46 0.06 1.07 2.05 1.75 -0.44 0.04 1.04 2.56 1.91 -0.45 0.08 1.07 2.37 1.81 -0.47 0.07 1.04 2.84 2.00 -0.44 0.06 1.11 2.79 1.92 -0.45 0.07 1.06 3.16 1.96 -0.42 0.05 1.09 3.23 1.93 -0.44 0.06 1.05 3.35 1.88 -0.43 0.05 1.07 3.68 1.97 -0.44 0.07 1.07 3.70 1.84 -0.45 0.06 1.08 4.07 1.92 -0.40 0.06 1.07 4.04 1.85 -0.44 0.06 1.07 4.37 1.95 -0.42 0.07 1.06 4.44 1.90 -0.47 0.11 1.08 4.72 2.02 -0.42 0.07 1.11 4.65 1.98 -0.44 0.07 1.08 5.10 1.98 -0.45 0.07 1.06 4.94 1.80 -0.46 0.04 1.07 5.36 1.98 -0.42 0.04 1.09 5.26 2.01 -0.43 0.07 1.10 5.86 1.80 -0.48 0.07 1.05 5.93 1.94 -0.44 0.08 1.05

Site 4 (Back-beach dune) Site 2 (Inland Dune) 1 0.57 1.71 -0.47 0.04 1.05 1 0.51 1.83 -0.46 0.05 1.05 1.18 1.74 -0.46 0.04 1.05 0.80 1.70 -0.47 0.04 1.05 1.11 1.67 -0.46 0.01 1.06 1.35 1.92 -0.44 0.07 1.07 1.65 1.75 -0.49 0.07 1.05 2 1.41 1.62 -0.46 0.02 1.08 1.93 1.87 -0.45 0.07 1.08 1.72 1.60 -0.45 0.02 1.07 2.00 1.45 -0.48 0.04 1.08 2.21 1.92 -0.44 0.07 1.08 2.54 1.84 -0.45 0.04 1.10 3 2.36 1.43 -0.50 0.01 1.04 2.83 1.81 -0.48 0.06 1.07 2.66 1.70 -0.48 0.04 1.05 3.09 1.83 -0.46 0.07 1.09 2.97 1.90 -0.44 0.05 1.07 3.38 1.80 -0.48 0.06 1.05 3.27 1.90 -0.46 0.05 1.06 3.91 1.80 -0.48 0.06 1.05 3.40 1.71 -0.51 0.05 1.03 4.21 1.84 -0.47 0.07 1.08 3.72 1.82 -0.47 0.04 1.05 4.50 1.75 -0.48 0.06 1.04 4.04 1.92 -0.47 0.06 1.05 4.77 1.71 -0.46 0.07 1.04 4.55 1.96 -0.44 0.05 1.05 5.07 1.93 -0.45 0.07 1.07 5.00 1.86 -0.49 0.06 1.03 5.31 1.88 -0.44 0.07 1.07 5.37 1.71 -0.47 0.02 1.04 5.54 1.81 -0.32 0.06 1.06 5.61 1.68 -0.45 0.01 1.07 5.85 1.67 -0.54 0.07 1.04 5.83 1.85 -0.46 0.04 1.07 6.22 1.82 -0.44 0.03 1.06 Site 5 (Inland dune) 1 0.68 1.83 -0.48 0.07 1.06 0.94 1.80 -0.47 0.06 1.06 1.23 1.71 -0.47 0.04 1.04 1.53 1.78 -0.49 0.06 1.04 1.78 1.90 -0.45 0.08 1.05 2.04 1.84 -0.48 0.06 1.07 2.30 1.92 -0.44 0.06 1.06 2.55 1.59 -0.47 0.01 1.07

This record was extended back to 350 BC and up to 283 storms by rainfall related flooding rather than sea inundation. Finally by the addition of the works of Galloway (2013), Kington the dataset was split into two: (1) storms where there was (2010), Zong and Tooley (2003), Harland and Harland (1980), an explicit reference to them causing coastal flooding in Lamb (1977) and others (see Table III) based on a variety eastern England (n = 189); and (2) storms where there was of archival documentary sources including floodstones major/catastrophic coastal flooding in NW Europe in which, (see http://floodstones.co.uk/), Parish records and the Times while the sources do not explicitly state flooding in England, Newspaper. Only storms where coastal flooding around the such flooding may have occurred (n = 21). As with all such North Sea had been reported were included. This storm record compilations of data, recording of storms is highly dependent was refined by filtering out pre 1200 AD events where the on them being deemed at the time sufficiently out of the record was limited (n = 12), storms reporting only minor ordinary to merit recording and/or have caused significant flooding not in Eastern England (n = 61) and summer storms damage/death. With increased coastal protection, coastal (n = 16). The latter was on the basis that these were not near flooding is likely to have diminished with time. Data compila- the Spring or Autumn equinox tides and more likely caused tions are also highly dependent on survival of the original

Figure 4. Down core stratigraphy and sample points of site 2 with particle size distributions from above, within and below what is interpreted as coarser aeolian unit deposited during a storm. Dashed red lines are the 5th and 95th percentiles and the orange solid line the 50th percentile. Note the shift to coarser sediment for the middle two samples, the extended coarse (left) tail and the greater width between the 5th and 95th percentile indicating poorer sorting. [Colour figure can be viewed at wileyonlinelibrary.com]

Table III. Main documentary sources used to compile coastal storm observed and recorded. Thus the compiled database can be flooding record. Also shown is the number of storms corroborated by expected to have an unquantified bias towards more storms other sources in the database being recorded more recently. A final caution stems from the judgement call of those compiling storm events (including the Period covered Number of Corroborated present authors) as to whether coastal flooding was entirely Documentary source (years AD) storms storms localised and thereby only of significance to the present study Brooks et al., 2016 1883–2013 20 8 if Norfolk was mentioned or more regional but only recorded Clemmensen at specific places. We can be reasonably sure that Norfolk et al., 2014 1872 1 was affected by the storm of 27 October 1882 as it caused Galloway, 2013 1287–1404 8 4 flooding to the south in London and to the north in Whitby Cunningham (Zong and Tooley, 2003). However, the storm of 21 December et al., 2011 1775–1176 2 1881 is only recorded as causing flooding in London (Zong – Kington, 2010 1099 2006 16 6 and Tooley, 2003) and it is unclear whether Norfolk was also Pye and Blott 2007 1851–1856 2 affected but not recorded in the London Times. Zong and Tooley, 2003 1788–1993 132 14 Harland and Harland, 1980 1287–1978 16 10 Results Lamb, 1977 245–1973 117 17 In this section the POSL, OSL dating and particle size analysis are initially looked at to establish the storm record as archived documents and, given their very diverse nature, finding them in the dunes. Major storm events are interpreted where both a for inclusion in the database. With the advent of the instrumen- phase change (step) occurs in the POSL profiles and there is ev- tal record, increased shipping and coastal activity and better idence for coarser sediments within a dune. Minor storms are record keeping so events are more likely to have been interpreted where there is evidence only for coarser sediments

© 2017 The Authors. Earth Surface Processes and Landforms published by John Wiley & Sons Ltd. Earth Surf. Process. Landforms, (2017) COASTAL SAND DUNES AND STORMS within a dune. The storm record interpreted from the dunes is resulting from three lower magnitude storms which caused then compared with the archived mapping of the area and dune accretion to briefly be interrupted with coarser sediment documented storm record. deposition before normal dune accretion resumed (similar to that shown in Figure 2(d)). The storm associated with the upper coarser layer at Site 4 dates to ~1981 ± 16 years AD. The Dune archive coarser layer at 2.0 m in site 1 probably corresponds with the 4.8 m layer in site 4 and the 1.65 m layer in Site 3 and this storm In the back-beach dunes (Sites 1, 3 and 4) POSL results show event dates to ~1966 ± 17 years AD. The coarser layer at 5.0 m that most have similar POSL signals all the way down the cores at site 3 is a third storm event which dates to after (Figure 5). These are interpreted as showing that the majority of 1943 ± 18 years AD. dune construction was essentially in a single phase (Figure 5). The inland dunes show a slightly more complex history. Site The only exceptions to this are the uppermost 1 m of sediments 5, based on the two phases separated by a step in the POSL at Site 1 and Site 4 which have a lower POSL signal. These are data (Figure 5), appears to show an upper sand unit 2.5 m thick, interpreted as indicating more recent small-scale dune modifi- which deposited in a single phase overlying a slightly older cation. The basal increase of POSL signal at Site 4 is believed sand unit. This is confirmed by the full OSL ages which suggest to relate to the underlying beach rather than the dune itself. Full the upper sediment accumulated around 1965 ± 20 years AD OSL ages support the POSL results. The dune at Site 4 dates to and the lower sediment at 1945 ± 20 years AD. Site 2 has three 1995 ± 30 years AD and the uppermost dune sediments at Site phases of accumulation recorded in the POSL data separated 1 dates to 1981 ± 16 years AD (dark brown shading in by two steps at around 1.4 m and 2.5 m. (Figure 5). An upper Figure 5). The lowermost dune sediments at Site 1 and the unit which full OSL ages date to 1938 ± 16 years AD and there- majority of dune sediments at Site 3 are attributed to around fore is of similar age to the basal sediments at Site 5 (light brown 1966 ± 17 years AD (mid-brown on Figure 5). The basal shading in Figure 5), a middle unit that unfortunately was not sediments at Site 3 show a phase change in the POSL and sampled for OSL dating and a lower unit dated to appear slightly older dating to 1943 ± 18 years AD (light brown 1874 ± 20 years AD (yellow shading in Figure 5). The inland shading in Figure 5). dunes appear to have ceased accumulating after around Particle size results for the back-beach sites showed coarser 1945 ± 20 years AD with dune formation switching to the layers at 2.0 m at Site 1, at 1.8 m, 5.0 m and 5.8 m at Site 3 present-day back beach dunes by 1966 ± 17 years AD and at 1.65 m and 4.8 m at Site 4 (see red stars in Figure 5). (Figure 6). Such a change in dunefield can likely be attributed One of these coarser layers, that at 5.8 m at site 3, corresponds to a significant storm event some time between 1945 and with a step up in POSL signal. This is interpreted as indicating a 1966. During this event sediment was likely eroded and moved major storm event that modified the dunes through either off-shore before being fed back onshore supplying sediment to erosion or dune movement and deposited a coarse layer before the present back beach dunes. dune accretion resumed after the storm. This major storm event Particle size data from the inland dune sites shows a coarser is dated at Site 3 to 1943 ± 18 years AD. The six other identified basal sediment at Site 5 and a coarser layer around 2.15 m coarse layers (with no step up in POSL) are interpreted as down the core at Site 2 (Table III). Both these coarser layers

Figure 5. Downcore portable OSL and mean grain size for sampled dune sites. Phases of dune accumulation based on phases within dunes identified by portable luminescence (red dashed lines) and between dunes based on OSL ages. Coarser sediments inferred to relate to storms indicated with red stars. Note sites have been re-ordered based on their age and localities with sites 1, 3 and 4 from the dune found at the back of the beach dunes and sites 2 and 5 from inland dunes. The background colour changes indicate sediments of a similar age based on the full OSL ages. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 6. Historical map and imagery evidence for Holkham progradation of shoreline in relation to the sampled dune sites. Note the seawall acts as a point of reference for all time slices shown. [Colour figure can be viewed at wileyonlinelibrary.com] correspond to changes in POSL and are interpreted as indicat- 1874–1945 years AD. A further two coarser layers are observed ing major coastal storm events which modified the dunes. at 3.4 m and 5.6 m depth at Site 2, which as the POSL does not The coarser layer at Site 5 probably corresponds to that found change significantly, are attributed to lower magnitude storms at Site 3 and dates the storm to around 1945 ± 20 years. The site which did not modify the dune and which occurred some time 2 coarser layer was not dated but falls within the periods after 1874 ± 20 years AD.

Table IV. Summary of the storms extracted from the dune archive at Holkham

Storm date Sites Dunes affected Evidence Storm magnitude Attributed storm

~1995 Site 4 Back-beach Coarse sand layer, no POSL Minor 1978 change ~1966 Sites 1,3,4 Back-beach Coarse sand layer, no POSL Minor 1961 change ~1945 Sites 3,5 Inland and Coarse sand layer, jump in Major 1953 Back-beach POSL and dune system avulsed to back-beach dunes ~1938 Site 2 Inland Dune system avulsed to Minor 1938 dunes associated with site 5 Between Site 2 Inland Coarse sand layer, jump in POSL Major 1897 ~1874–1945 Between Site 2 Inland Coarse sand layer, no POSL Minor 1883 ~1874–1945 change Between Site 2 Inland Coarse sand layer, no POSL Minor 1875 ~1874–1945 change Pre 1874 Site 2 Inland Inland dune initiation and Major 1850’s? presumed dune avulsion from previous dunes

In summary, the dunes sampled provide a ~ 140 year record In a reanalysis of synoptic pressure patterns using the (Table IV). Within this, at least two major storm events modified Jenkinson daily weather type for the Lincolnshire coast, the dunes and/or caused evulsion of the system and new dune- Montreuil and Bullard (2012) also reconstructed on-shore lines to be formed. One took place in the late 19th to early 20th storms for the period 1870–2012. They found that wind storms century and a second in the 1950s. Five smaller storms are were higher than average in terms of frequency between 1871 interpreted from the record to have occurred in the 1980s, and the 1920s, relatively infrequent up to until 1955 and 1960s, 1930s and two late in the 19th century. peaked 1955–1970, 1975–1988 and from the mid-1990s to 2000. This closely follows the documentary based storm record.

Archive mapping and storm record Discussion

Historical maps and imagery show that the inland dune sites Documentary evidence and luminescence dating have rarely (sites 2 and 5) could not have existed prior to 1840 as where been combined in an integrated analysis (Benito et al., 2011). they were found was part of the beach before this time This study is the first to attempt it on coastal systems. Based on (Figure 6). Coastal progradation after this date would have the luminescence and particle size data the Holkham dune sed- provided accommodation space for the formation of the dunes imentary archive has recorded two major storm events (one late associated with Site 2. Maps show that site 1, 3 and 4 (presently 19th–early 20th century, one in the 1950s) of sufficient magni- at the back of the beach) only appear above the high-tide line tude to modify dunes and/or cause evulsion of the system and after 1970 and first appears on an aerial photograph from a further five other storms (two late 19th century, 1930s, 1960s 1988. Within the limits of the time slices available this confirms 1980s). Clearly the dune archive is not preserving a record of that over the last 160 years the coastline associated with the all the documented storms within this period (Figure 7(b)). Holkham dunes has been prograding northwards. The mapping Based on documentary evidence for the last 140 years, an aver- shows some erosion of dunes on the western side of the study age of three storms a decade affect East Anglia compared with area during the 1930s and 1950s as indicated by Site 4 being the 0.5 storms per decade extracted from the Holkham dune ar- further offshore at these points. A significant re-organisation chive. That the dunes at Holkham are not recording all docu- appears to have taken place between 1973 and 1988 whereby mented storms in the region probably in part reflects that not a formerly inter-tidal and beach area rapidly became dunes. all the documented storms impacted directly on Holkham. On The map/aerial imagery supports both the full SAR OSL ages average only 2.2 storms per decade are recorded if storms are and the POSL data. restricted to only those documented as directly impacting on Over the last c. 800 years, the record reconstructed from Norfolk. While reducing the mismatch, this still probably over- observations and documentary sources indicates that storms represents the number of storms directly experienced at Holkham. causing coastal flooding in eastern England occurred approxi- The key to preserving a storm trace in the dune archive is the mately once every nine years (Figure 7(a)). Relatively low storm magnitude of the storm and any accompanying storm surge. frequency is apparent within the period 1400–1600 AD and a Some storms simply will not be of sufficient magnitude to cause very large apparent increase in storminess has occurred since dune erosion and/or deposition of a coarse sand layer on the 1800. As previously discussed this can at least partly be attrib- crest. Given the age ranges of the dated levels in the dune ar- uted to better and more frequent records. Focusing on the chive, an exact attribution to known high magnitude storm period since 1800 and the East Anglia region (Figure 7(b)), once surges cannot be given. However, in the recent record, the last again storms causing coastal flooding appear to have increased significant flooding event prior to 2013 was on 11 January with time, averaging over the whole period to two per decade. 1978 (Figure 8; Table IV). During this storm surge extensive Of note are the decades of 1870s, 1900s, 1950s, 1970s, 1990s flooding occurred across north Norfolk (Harland and Harland, and 2000s, decades that all seem to have had four or more 1980; Brooks et al., 2016). This may have laid down the up- storms in East Anglia. There is no correlation between permost coarse sand layers identified in the back-beach dunes. frequency of flooding in East Anglia and that of eastern England The storm identified as having deposited a coarser layer in the (correlation coefficient = À0.074). back beach dunes during the 1960s is less certain. Harland

Figure 7. Storm record compiled from published records (see text for details). (a) Storm record since 1200 AD (number of storms per century) including those reported for eastern England and major storm events elsewhere in southern North Sea known to have cause significant flooding/loss of life. (b) Storm record back to 1800 AD (number of storms per decade) for East Anglia and elsewhere in eastern England. Blue arrows denote the approx- imate age when the Holkham dune archive indicates that a new line of dunes formed. [Colour figure can be viewed at wileyonlinelibrary.com] and Harland (1980) describe the 20 March 1961 storm as post a major coastal storm event (as per the 1953 flooding) causing ‘heavy seas, strong winds and damage and destruc- then the establishment of the inland dunes around site 2 must tion’ but no sea elevation data is known. More certain is the have been before ~1878. The line of inland dunes could 1953 storm surge which attained a sea elevation of 5.15 m therefore relate to dune avulsion after the series of powerful OD at Wells-Next-The-Sea (Brooks et al., 2016; Figure 7) storms in the 1850s which are recorded as having breached and caused major flooding in the area (Figure 8). It would Spurn Head near Hull and caused coastal flooding from Hull, seem likely that the coarser layers with associated erosion the Fenlands down to London (Zong and Tooley, 2003). found in Sites 3 and 5 along with the change in dune position from the inland dunes (before 1950s) to the present-day back beach dunes (post 1960; Figure 5) relates to this major event Future Directions (Table IV). In a study of the UK’s Lincolnshire coastline, Montreuil and Bullard (2012) noted the symbiotic relationship Overall, from the above it would appear that, with the appro- of storms initially causing coastal erosion but then transporting priate interrogation, coastal dunes can act as sedimentary back onshore large quantities of sediment to reform dunes. 15 archives of storm surge activity. If such research ultimately aims February 1938 saw a storm surge which flooded the to give coastal managers a better understanding of return inter- Brancaster marshes and the contours of beach and sandhills vals of coastal storm surges and spatial extent of flooding then (dunes) completely changed (Harland and Harland, 1980). moving beyond the instrumental record is required. They also Sea elevations during the storm surge are not known. This show that while dune erosion/breaching during major storms may be recorded in the sand dune archive as the upper most may be dramatic, providing they are followed by a number of sediment unit at Site 2 after which mapping shows the dunes years without such storms, dunes appear to recover by either avulsed to those associated with site 5 (Figure 6). The rebuilding or re-establishing a new dune line. At Holkham, ex- major storm event associated with the middle sand unit at tending the record to older inland dunes should uncover Site 2 could relate to the flood associated with the storm of sedimentary archives for documented storms in March 1820, 28 November 1897 (Figure 8) in which flooding reached November 1794, February 1736, January 1734 and even 4.49 m OD (Brooks et al., 2016; Table IV). The minor storms potentially November 1404 (Kington, 2010, Harland and identified in the lower Site 2 dune sediments could be from Harland, 1980, Galloway, 2013). All these are reported to have 11 March 1883 storm which flooded to 5.49 m OD at Wells- had major storm surges associated with them (ibid). Next-The-Sea (ibid) and the storm of 14 October 1875 which Research on the dune sedimentary archive will only really flooded coasts from London to Whitby (Zong and Tooley, add value to coastal managers when it is able to, with confi- 2003; Table IV). While undated, if dune-line initiation occurs dence, identify previously unrecorded storm surges at localities

• An alternative explanation for the difference between documented storms and the Holkham dunes is that only high magnitude events erode and/or deposit sand on dunes and so it is only high magnitude events that are recorded in the dune archive. Further research would be needed to establish a better idea of a threshold above which storm events sufficiently impact on dunes for them to be recorded in the sedimentary record. Such an approach has been success- fully undertaken in the context of the relationship between beach storm ridges and tropical cyclone activity in Australia (Forsyth et al., 2010).

Conclusions

Application of POSL profiling with supporting luminescence ages and detailed particle size analysis to coastal dunes, as well as providing information of dunefield evolution can also provide information of past coastal storms. In this study, seven storm events, two major, were identified from the dune archive. These spanned the last 140 years and seemed to fit well with documentary evidence of major storm surges which have affected north Norfolk. Targeting of dunefields, which stack sediment rather than overturning, greatly enhances the potential for dunes to retain records of storm events and for successful application POSL profiling. Dunes do not appear to be recording (at least at the sampling resolution used here) all recorded storm events only ones in which sea elevations were particularly elevated and significant flooding occurred. As such the approach seems to hold promise to extend the dune archive records further back in time when less is known about storm surges to get a better understanding of their frequency, and older dunes and extending the record to a regional level would Figure 8. Preserved flood markers around North Norfolk. (a) Flood allow this. The work also has coastal management implications markers for 2013 at 6.31 m OD, 1978 at 5.94 m OD and 1953 at in that, providing high intensity storm events are followed by a 6.15 m storms found on the Granary building, Wells-Next-The-Sea. (b) Flood markers for 1978 at 5.93 m OD and 1953 at 6.13 m OD storms number of years without such storms, dunes recover by either found on Beach Road Wells-Next-The-Sea. (c) Flood markers for 1953 rebuilding or re-establishing a new dune line. at 6.55, OD, 1978 at 5.98 m OD and 1897 at 5.54 m OD storms recorded on the Quay building in Blakeney. [Colour figure can be Acknowledgements—Sarah Henderson and the Holkham Estate are viewed at wileyonlinelibrary.com] thanked for their assistance in working at Holkham. Gill Tyson is thanked for her cartographic assistance with the figures. This research was funded through a White Rose Consortium Grant held by and/or extend the record back to where documentary evidence Dr Katherine Selby at the University of York. It was also funded through is patchy. Before this is achieved further evaluative steps should a NERC grant to GR (Ne/L002450/1). be undertaken:

• A better understanding is required as to whether subsequent References events could, through erosion, ‘overwrite’ earlier storm archives and closely spaced events might be combined in Andrews JE, Boomer I, Bailiff I, Balson P, Bristow C, Chroston PN, the dune archive. This may account for the decades Funnell BM, Harwood GM, Jones R, Maher BA, Shimmield GB. where four or more storm events have been documented 2000. Sedimentary evolution of the north Norfolk Barrier coastline in the context of Holocene sea level change. In Holocene Land- (e.g. 1900s and 1950s, Figure 7(b)), but only one or less Ocean Interaction and Environmental Change around the North events are recorded in the dunes. Use of ground penetrating Sea, Shennan I, Andrews J (eds), Vol. 166. Geological Society, radar, as employed by Cunningham et al. (2011), and London, Special Publications: London; 219–251. continuous or more high resolution sampling for POSL might Bateman MD, Catt JA. 1996. An absolute chronology for the raised help in this regard. beach and associated deposits at Sewerby, East Yorkshire, England. • That the dunes at Holkham are not recording all docu- Journal of Quaternary Science 11: 389–395. mented storms in the region also probably reflects that not Bateman MD, Stein S, Ashurst RA, Selby K. 2015. Instant luminescence all these storms impacted directly on Holkham. Sixteen chronologies? High resolution luminescence profiles using a portable coastal dune fields exist for Norfolk and Suffolk (Pye et luminescence reader. Quaternary Geochronology 30: 141–146. – al., 2007). Future research extending the examination of Baxter PJ. 2005. The East Coast Big Flood, 31 January 1 February 1953: a summary of the human disaster. Philosophical Transactions of the the dune archive to this extensive regional dune record Royal Society of London 363A: 1293–1312. would remove the current mismatch between comparing a Benito G, Thorndycraft VR, Rico MT, Sánchez-Moya Y, Sopeña A, regional documentary storm record with a site-specific Botero BA, Machado MJ, Davis M, Pérez-González A. 2011. Hydro- dune archive record. This would also allow a more rigorous logical response of a dryland ephemeral river to southern African evaluation of the efficacy of the dune archive in recording climatic variability during the last millennium. Quaternary Research storm events. 75: 471–482.

Bristow CS, Chroston NP, Bailey SD. 2000. The structure and develop- Oronsaye WI. 1990. Wind action and sand movement near Holkham ment of foredunes on a locally prograding coast: insights from Bay, North Norfolk Coast, England. Environmental Geology and ground-penetrating radar surveys, Norfolk, UK. Sedimentology 47: Water Sciences 15:77–82. 923–944. Puijenbroek MEB, Limpens J, deGroot AV, Riksen MJPM, Gleichman Brooks SM, Spencer T, McIvor A, Möller I. 2016. Reconstructing and M, Slim PA, van Dobben HF, Berendse F. 2017. Embryo dune understanding the impacts of storms and surges, southern North development drivers: beach morphology, growing season precipita- Sea. Earth Surface Processes and Landforms 41: 855–864. tion and storms. Earth Surface Processes and Landforms 42: Church JA, Clark PU, Cazenave A, Gregory JM, Jevrejeva S, Levermann 1733–1744. A, Merrifield MA, Milne GA, Nerem RS, Nunn PD, Payne AJ, Pfeffer Pye K, Blott, S. 2007. Holderness Erosion and Evolution of the Spurn WT, Stammer D, Unnikrishnan AS. 2013. Sea level change. In Peninsula. Unpublished report. http://www2.hull.ac.uk/science/ Climate Change 2013: The Physical Science Basis, Stocker TF, Qin PDF/ECSA%20Ken%20Pye.pdf [last accessed 9/5/17] D, Plattner G-K, Tignor M, Allen SK, Boschung J, Nauels A, Xia Y, Pye K, Saye SE, Blott SJ. 2007. Sand dune processes and management Bex V, Midgley PM (eds). The Fifth Assessment Report of the Intergov- for flood and coastal defence. Parts 1–5. Joint DEFRA/EA Flood and ernmental Panel on Climate Change. Cambridge University Press: Coastal Erosion Risk Management R and D Programme, R and Cambridge, UK and New York, NY, USA. D Technical Report FD1302/TR. DEFRA/EA, London, 5 volumes. Clemmensen LB, Bendixen M, Hede MU, Kroon A, Nielsen L, Murray Sanderson DW, Murphy S. 2010. Using simple portable OSL measure- AS. 2014. Morphological records of storm floods exemplified by ments and laboratory characterisation to help understand complex the impact of the 1872 Baltic storm on a sandy spit system in and heterogeneous sediment sequences for luminescence dating. south-eastern Denmark. Earth Surface Processes and Landforms 39: Quaternary Geochronology 5: 299–305. 499–508. https://doi.org/10.1002/esp.3466 Shennan I, Horton BP. 2002. Holocene land- and sea-level changes in Cunningham AC, Bakker MAJ, van Heteren S, van der Valk B, van der Great Britain. Journal of Quaternary Science 17:511–526. Spek AJF, Schaart DR, Wallinga J. 2011. Extracting storm-surge data Sibley A, Cox D, Titley H. 2015. Coastal flooding in England and Wales from coastal dunes for improved assessment of flood risk. Geology from Atlantic and North Sea storms during the 2013/2014 winter. 39: 1063–1066. Weather 70:62–70. East Anglian Coastal Group. 2010. North Norfolk Shoreline Manage- Spencer T, Brooks SM, Evans BR, Tempest JA, Möller I. 2015. ment Plan Final Plan. http://www.eacg.org.uk/docs/smp5/the% Southern North Sea storm surge event of 5 December 2013: water 20smp%20main%20report.pdf. [last accessed 9/5/17] levels, waves and coastal impacts. Earth Science Reviews 146: Forsyth AJ, Nott J, Bateman MD. 2010. Beach ridge plain evidence of a 120–145. variable late-Holocene tropical cyclone climate, North Queensland, Steers JA, Stoddart DR, Bayliss-Smith TP, Spencer T, Durbidge PM. Australia. Palaeogeography, Palaeoclimatology, Palaeoecology 297: 1979. The storm surge of 11 January 1978 on the east coast of 707–716. England. The Geographical Journal 145: 192–205. Galloway JA. 2013. Coastal flooding and socioeconomic change in Vousdoukas MI, Voukouvalas E, Annunziato A, Giardino A, Feyen L. eastern England in the later middle ages. Environment and History 2016. Projections of extreme storm surge levels along Europe. 19: 173–207. Climate Dynamics 47: 3171–3190. Gerritsen H. 2005. What happened in 1953? The Big Flood in the Wolf J, Flather RA. 2005. Modelling waves and surges during the 1953 Netherlands in retrospect. Philosophical Transactions of the Royal storm. Philosophical Transactions of the Royal Society of London Society London 363A: 1271–1291. Series A 363: 1359–1375. Harland MG, Harland HJ. 1980. The Flooding of Eastern England. Zappa GZ, Shaffrey LC, Hodges KI, Sansom PG, Stephenson DB. 2013. Minimax Books: Norwich. A multimodel assessment of future projections of North Atlantic Hiskey C. 2003. The Drainage of Holkham Marshes. Unpublished extratropical cylclones in the CMIP5 climate models. Journal of report for Holkham Estate Archives. Climate 26: 5846–5862. Kington JA. 2010. Climate and Weather. Collins: London. Zong Y, Tooley MJ. 2003. A historical record of coastal floods in Lamb HH. 1977. Climate Past Present and Future. Volume 2, Climatic Britain: frequencies and associated storm tracks. Natural Hazards History and the Future. Methuen and Co: London. 29:13–36. Lancaster N. 2013. Geomorphology of Desert Dunes.Routledge:Oxford. Lewis M, Schumann G, Bates P, Horsburgh K. 2013. Understanding the variability of an extreme storm tide along a coastline. Estuarine, Supporting Information Coastal and Shelf Science 123:19–25. Montreuil A-L, Bullard JE. 2012. A 150-year record of coastline dynam- Additional Supporting Information may be found online in the ics within a sediment cell: Eastern England. Geomorphology 179: supporting information tab for this article. 168–185. Murray AS, Wintle AG. 2003. The single aliquot regenerative dose Figure S1. Example radial plots of OSL palaeodoses (De) for the protocol: potential for improvements in reliability. Radiation three samples from site 2 (left column) and the two samples Measurements 37: 377–381. from site 5 (right column).